How? Near enough no one was using the Kunrei system for any of that. If anything this will make it more consistent or at least no worse. Macrons are the biggest inconsistency but that’s always been the case.
It was either Hepburn, the English title (i.e. rock instead of rokku), or just most sensibly kana/kanji that would have been used for this everywhere, never other romanisation systems, to within a rounding error.
It was almost never quite Hepburn either, usually shi/chi/tsu/fu/ji with no di/du, but often alongside wo/he/ha (in roughly that order of likelihood, not always consistently), macrons almost never, っち is cch. Ironically, I have to imagine there's more "bastardized Nihonsiki" out there than "bastardized Kunreisiki", because the differences between the two are exactly the ones that matter when typing them out, and of course everyone in the j/e scenes is by far most often inputting wa-puro ro-maji (and of course that's ji, not zi, because which one is on the home row?).
In short, the usual infelicities of Japanese romanization as practiced in the wild on keyboards people actually have, and there is a method to the madness but it's not what any of the standards reflect.
It was either Hepburn, the English title (i.e. rock instead of rokku), or just most sensibly kana/kanji that would have been used for this everywhere, never other romanisation systems, to within a rounding error.